Domain adaptation of artificial intelligence models based on multi-source time-series data

Abstract

Systems and methods for domain adaptation of artificial intelligence (AI) models based on multi-source time-series data. Meta-data information from tuples of time-series data and corresponding labels for the time-series data can be learned based on a fidelity loss with a prompt-based deep learning model (POND) using determined soft prompts. Mutual information from the meta-data information can be minimized by minimizing a discrimination loss of domain-specific information from the meta-data information. A common prompt can be learned with a learning objective that combines a training loss, the fidelity loss and the discrimination loss. AI models can be adapted to perform downstream tasks for different domains by utilizing the common prompt for the AI models.

What is claimed is: 
     
         1 . A computer-implemented method, comprising:
 learning meta-data information from tuples of time-series data and corresponding labels for the time-series data based on a fidelity loss with a prompt-based deep learning (POND) model using determined soft prompts;   minimizing mutual information from the meta-data information by minimizing a discrimination loss of domain-specific information from the meta-data information;   learning a common prompt with a learning objective that combines a training loss, the fidelity loss and the discrimination loss; and   adapting artificial intelligence (AI) models to perform downstream tasks for different domains by utilizing the common prompt for the AI models.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein the downstream tasks further comprises updating a medical diagnosis of a patient from healthcare data collected as time-series data by utilizing the AI models to assist a decision-making process of a decision-making entity. 
     
     
         3 . The computer-implemented method of  claim 1  wherein learning the meta-data information further comprises determining soft prompts that preserve meaningful information between time-series input data and meta-data information. 
     
     
         4 . The computer-implemented method of  claim 3 , wherein determining the soft prompts further comprises maximizing the mutual information between a generated prompt and a corresponding label for the time-series input data. 
     
     
         5 . The computer-implemented method of  claim 4 , wherein learning the meta-data information further comprises computing the fidelity loss using predictions of the POND model from a concatenation of time-series data and the generated prompt. 
     
     
         6 . The computer-implemented method of  claim 1 , wherein learning the common prompt further comprises optimizing a prompt generator in a target domain using time-series data and corresponding labels for the target domain with few-shot transfer. 
     
     
         7 . The computer-implemented method of  claim 1 , wherein learning the common prompt further comprises employing meta learning for the POND model to learn the learning objective. 
     
     
         8 . A system, comprising:
 a memory device;   one or more processor devices operatively coupled with the memory device to perform operations including:
 learning meta-data information from tuples of time-series data and corresponding labels for the time-series data based on a fidelity loss with a prompt-based deep learning (POND) model using determined soft prompts; 
 minimizing mutual information from the meta-data information by minimizing a discrimination loss of domain-specific information from the meta-data information; 
 learning a common prompt with a learning objective that combines a training loss, the fidelity loss and the discrimination loss; and 
 adapting artificial intelligence (AI) models to perform downstream tasks for different domains by utilizing the common prompt for the AI models. 
   
     
     
         9 . The system of  claim 8 , wherein the downstream tasks further comprises updating a medical diagnosis of a patient from healthcare data collected as time-series data by utilizing the AI models to assist a decision-making process of a decision-making entity. 
     
     
         10 . The system of  claim 8 , wherein learning the meta-data information further comprises determining soft prompts that preserve meaningful information between time-series input data and meta-data information. 
     
     
         11 . The system of  claim 10 , wherein determining the soft prompts further comprises maximizing the mutual information between a generated prompt and a corresponding label for the time-series input data. 
     
     
         12 . The system of  claim 11 , wherein learning the meta-data information further comprises computing the fidelity loss using predictions of the POND model from a concatenation of time-series data and the generated prompt. 
     
     
         13 . The system of  claim 8 , wherein learning the common prompt further comprises optimizing a prompt generator in a target domain using time-series data and corresponding labels for the target domain with few-shot transfer. 
     
     
         14 . The system of  claim 8 , wherein learning the common prompt further comprises employing meta learning for the POND model to learn the learning objective. 
     
     
         15 . A non-transitory computer program product comprising a computer readable storage medium including program code for domain adaptation of artificial intelligence (AI) models based on multi-source time-series data, wherein the program code when executed on a computer causes the computer to perform operations having:
 learning meta-data information from tuples of time-series data and corresponding labels for the time-series data based on a fidelity loss with a prompt-based deep learning (POND) model using determined soft prompts;   minimizing mutual information from the meta-data information by minimizing a discrimination loss of domain-specific information from the meta-data information;   learning a common prompt with a learning objective that combines a training loss, the fidelity loss and the discrimination loss; and   adapting AI models to perform downstream tasks for different domains by utilizing the common prompt for the AI models.   
     
     
         16 . The non-transitory computer program product of  claim 15 , wherein the downstream tasks further comprises updating a medical diagnosis of a patient from healthcare data collected as time-series data by utilizing the AI models to assist a decision-making process of a decision-making entity. 
     
     
         17 . The non-transitory computer program product of  claim 15 , wherein learning the meta-data information further comprises determining soft prompts that preserve meaningful information between time-series input data and meta-data information. 
     
     
         18 . The non-transitory computer program product of  claim 17 , wherein determining the soft prompts further comprises maximizing the mutual information between a generated prompt and a corresponding label for the time-series input data. 
     
     
         19 . The non-transitory computer program product of  claim 18 , wherein learning the meta-data information further comprises computing the fidelity loss using predictions of the POND model from a concatenation of time-series data and the generated prompt. 
     
     
         20 . The non-transitory computer program product of  claim 15 , wherein learning the common prompt further comprises optimizing a prompt generator in a target domain using time-series data and corresponding labels for the target domain with few-shot transfer.